Anthropic publishes Opus 4.7 prompts
- Anthropic said on April 16 that Claude Opus 4.7 is now generally available across Claude, its API, Amazon Bedrock, Google Vertex AI and Microsoft Foundry. - Anthropic lists Opus 4.7 at a 1 million-token context window and $5 per million input tokens, while positioning it as a step up from Opus 4.6 for agentic coding. - The release also ties Opus 4.7 to new cyber-use safeguards and Anthropic’s broader transparency push around model deployment and risk controls. (anthropic.com)
Anthropic said on April 16 that Claude Opus 4.7 is now generally available, making its newest flagship model available across Claude, its developer API and major cloud partners. (anthropic.com) The company said Opus 4.7 improves on Opus 4.6 in advanced software engineering, especially on harder coding tasks that previously needed closer human supervision. (anthropic.com) Anthropic’s model docs list a 1 million-token context window for Opus 4.7, with pricing of $5 per million input tokens and $25 per million output tokens. (docs.claude.com) (anthropic.com) The same docs say Opus 4.7 uses adaptive thinking, not the older manual extended-thinking setting, and supports a maximum output of 128,000 tokens. (docs.claude.com 1) (docs.claude.com 2) Anthropic is framing the model less as a chatbot update than as an engine for coding agents, long-running tasks and enterprise workflows that need sustained attention over large codebases. (anthropic.com 1) (anthropic.com 2) The release also sits inside Anthropic’s April 2026 cybersecurity push. In its launch post, the company said Opus 4.7 is the first model released with automatic safeguards meant to detect and block prohibited or high-risk cyber requests. (anthropic.com) Anthropic said security professionals can apply to a new Cyber Verification Program for legitimate uses such as vulnerability research, penetration testing and red-teaming. (anthropic.com) That matters because Anthropic has kept its more capable Claude Mythos Preview restricted to invited partners for defensive cybersecurity work, using Opus 4.7 as the broader public test case for those safeguards. (anthropic.com 1) (anthropic.com 2) Anthropic has published system cards and a Transparency Hub for recent Claude models, but the company’s official April 16 materials focus on Opus 4.7’s release, pricing, benchmarks and safety controls rather than a separate announcement centered on publishing its full system prompt. (anthropic.com) (anthropic.com) (anthropic.com) So the clearest story is not a standalone “prompt dump.” It is Anthropic shipping Opus 4.7 as a pricier, longer-context coding model while using it to roll out new cyber guardrails in public. (anthropic.com) (docs.claude.com)
